Central Circuitry in the Jellyfish Aglantha Digitale

The Journal of Experimental Biology 198, 2261Ð2270 (1995) 2261 Printed in Great Britain © The Company of Biologists Limited 1995

CENTRAL CIRCUITRY IN THE JELLYFISH AGLANTHA DIGITALE

I. THE RELAY SYSTEM

G. O. MACKIE1 AND R. W. MEECH2 1Biology Department, University of Victoria, Victoria, British Columbia, Canada and 2Department of Physiology, University Walk, Bristol, UK

Accepted 9 June 1995

Summary 1. The relay system is an interneuronal pathway in the 3. The relay system can conduct impulses on its own but margin of the jellyfish Aglantha digitale. It excites a second their conduction velocity is greatly increased when interneuronal pathway, the carrier system, and is itself preceded by either pacemaker or ring giant spikes. This excited by pacemaker neurones concerned with slow phenomenon, termed the ‘piggyback effect’, may be due to swimming. It also excites a slow conduction pathway in the extracellular field effects rather than to actions mediated tentacles causing graded, tonic contractions of all the by chemical or electrical synapses. tentacles during slow swimming. 4. Recordings from the epithelial cells that ensheath the 2. The pacemakers, the carrier system and the relay ring giant and outer nerve ring neurones show miniature system all contribute to the production of excitatory synaptic potentials and other events that seem to reflect postsynaptic potentials (EPSPs) in a giant axon that runs events in the nervous system, but no functions can be in the outer nerve ring (ring giant axon). These EPSPs may assigned to them. cause the latter to spike during slow swimming. If it does 5. There is no obvious counterpart to the relay system in so, it will fire tentacle giant axons, producing twitch medusae lacking escape circuitry. contractions of the tentacles. Such contractions probably help to contract the tentacles rapidly at the start of slow Key words: Cnidaria, jellyfish, hydromedusan behaviour, escape swimming. This is an unusual case of a giant axon that swimming, nervous system, nerve ring, giant axon, intracellular normally mediates escape behaviour being appropriated recording, relay system, carrier system, pacemaker, central circuitry, for use during a non-escape activity. Aglantha digitale.

Introduction The existence of bundles of neurones in the margins of evidence from electron microscopy, immunocytochemistry, dye hydromedusae has been known since 1850 when Louis Agassiz injection and extra- and intracellular recordings. described them in Bougainvillea superciliaris (see Mackie, We are now in a position to start filling in the picture for 1989a), and the physiological analysis of hydromedusan another jellyfish, Aglantha digitale. Along with other members behaviour, launched by Romanes (1876), has continued to the of the Family Rhopalonematidae (Mills et al. 1985), this present day. It was Passano (1965) who first applied medusa has neuronal pathways specialized for escape electrophysiological techniques to hydromedusae, uncovering a swimming and rapid, twitch-type tentacle contractions in wealth of conduction pathways and pacemaker systems in the addition to systems driving normal ‘slow’ swimming and slow, marginal nerve bundles. These ‘nerve rings’ are now graded tentacle responses. Escape responses are conducted understood to be the animal’s central nervous system in which around the margin by a ‘ring giant’ axon located in the outer all the major pathways interface and interact synaptically as in nerve ring (Roberts and Mackie, 1980). The ring giant elicits the ganglia of higher invertebrates. As the physiological large excitatory postsynaptic potentials in eight ‘motor giant’ analysis proceeded, it became possible to devise increasingly axons located in the subumbrellar myoepithelium (Meech and comprehensive wiring diagrams of the central circuitry in forms Mackie, 1995). EPSPs that exceed threshold give rise to a large such as Stomotoca atra (Mackie and Singla, 1975; Mackie, overshooting sodium-dependent action potential which is 1975) and Polyorchis penicillatus (Spencer, 1978, 1979; rapidly conducted to the swimming muscles, causing a strong Spencer and Arkett, 1984; Arkett and Spencer, 1986). The escape contraction. The ring giant also excites giant axons in Polyorchis work has gone furthest in this direction, synthesizing the tentacles (‘tentacle giants’), causing them to contract rapidly 2262 G. O. MACKIE AND R. W. MEECH during the escape response. The anatomy of the central nervous specialized escape circuitry. By providing a pathway linking system of A. digitale, based on earlier histological studies the neurones responsible for escape and non-escape swimming, (Roberts and Mackie, 1980; Mackie, 1989a,b; Mackie et al. they allow the animal to draw upon an escape component in 1989; Bickell-Page and Mackie, 1991), is depicted in Fig. 1. its normal swimming behaviour. Slow swimming is initiated by rhythmically active pacemaker neurones in the inner nerve ring. These pacemaker neurones also excite the motor giant axons but, unlike the ring Materials and methods giant, their slower, low-amplitude EPSPs (Meech and Mackie, Specimens of Aglantha digitale Müller were caught off the 1995) give rise to a calcium-based action potential that dock at the Friday Harbor Laboratories, Washington, USA, propagates relatively slowly to the swimming muscles and during May, June and July and were kept in the laboratory at produces a weaker muscle response. Thus, the ability of the 7 ûC. They were dissected in sea water containing about motor giants to conduct two sorts of action potential (Mackie 115 mmol lϪ1 Mg2+, which relaxes the tentacles and prevents and Meech, 1985) is matched by the existence of two sorts of swimming. As even small A. digitale (height 0.9 cm) may have EPSP (Meech and Mackie, 1995). The role of a family of as many as 80 tentacles, each up to 7 mm long, the tentacles potassium channels in regulating the excitability of the axon were cut short close to their bases to prevent them from getting membrane so as to allow separate sodium and calcium spikes in the way of the recording electrodes. Preparations, which has been explored by Meech and Mackie (1993a,b). consisted of a section of margin about 0.6 cm long (one-quarter Each of the two clearly defined forms of swimming has its of the circumference of the animal) and about 0.4 cm in the own specialized neuronal pathway, but the two systems are not radial direction, were pinned out in small Petri dishes lined completely independent and we show in this paper that subtle with Sylgard 184 (Dow-Corning Corp.) using spines of the interactions occur between them. A key mediator of these cactus Opuntia. The most useful spines are about 0.5 mm long. interactions, and the main focus of the present paper, is a ring Glass microelectrodes were used to cut or isolate nerves by of interneurones we call the ‘relay’ system because it relays scoring the neighbouring myoepithelium. Experiments were excitation from the pacemaker neurones to the ring giant axon. conducted in sea water unless otherwise stated. Another interneuronal system involved in exciting the ring Extracellular recordings were made with suction electrodes giant, the ‘carrier’ system, will be discussed in the companion pulled from heated polyethylene tubing. The amplified signals to this paper (Mackie and Meech, 1995). These systems have were filtered and displayed on a storage oscilloscope equipped no known counterparts in hydromedusae that lack the with a wave-form digitizer. They were recorded on Polaroid

Fig. 1. (A) Outline drawing of a A medium-sized Aglantha digitale, 1.2 cm bell height. The proportions change somewhat as the animal grows. (B) Perradial B section through the area boxed in A to show marginal nerve centres. The inner and outer nerve rings are bundles of neurones running in the Mesogloea ectoderm at the base of the velum. There are about 800 neurones in a typical cross section, most of them less than Motor giant 1.0 ␮m in diameter. The motor Tentacle giant, ring giant and tentacle Endoderm giant giant axons are conspicuously larger than other neurones Inner nerve ring running in the same regions. Both the ring giant and the Velum tentacle giants have prominent Tentacle central vacuoles. Nerves are shown which cross the Outer nerve ring Hair mesogloea at the base of the Pad Ring cell velum and connect the outer and cell giant inner nerve rings. The basal processes of the hair cells make contact with the ring giant axon directly. The pad cell sends a long process under the outer nerve ring, contributing to the epithelial ensheathment of groups of neurones. To simplify the picture, the connections passing around the sides of the tentacle that connect the tentacle giant axon to the outer nerve ring have been omitted along with the small nerve bundle that runs beside the motor giant and the small-diameter nerve plexuses running in the outer velar ectoderm and tentacle ectoderm. The relay system in a jellyﬁsh 2263

film. Most events recorded extracellularly were captured A within the 1 Hz to 1 kHz waveband. Signals were also recorded on an instrumentation tape recorder and/or a chart recorder. Unless otherwise stated, negative potentials are in the upward direction in the extracellular records. It should be noted that an Motor giant electrode on one side of the margin can pick up events from axon both nerve rings. For intracellular recordings, glass capillary micropipettes, RU Ϫ1 filled with 3 mol l potassium chloride, were mounted on Inner nerve ring Zeiss (Jena) manipulators fitted with magnetic bases. Amplification of the intracellular signal was conventional and Velum provided for intracellular current injection via a bridge circuit. Stimuli were also applied externally through small co-axial Stimulus bipolar metal electrodes (Clark Electromedical Instruments, RL SNE-100). The preparation was viewed with a stereoscopic binocular BC microscope mounted over a special transillumination base incorporating a double mirror system (Mackie, 1976a). Preparations were kept cool (10Ð12 ûC) during recordings using a Cambion thermoelectric cooling unit with a doughnut- shaped cooling platform that allowed transillumination of the 20 mV specimen during recordings. * P R W Elevated Mg2+ concentrations, used to reduce muscle 200 µV 20 ms contractions and epithelial after-potentials associated with . nervous activity, were obtained by adding 7 % MgCl2 6H2O to Fig. 2. (A) A drawing of the preparation pinned with its subumbrellar the sea water containing the preparation. Octanol and heptanol side up. An intracellular recording electrode (RU) is shown in the were used at concentrations between 0.2 and 1.0 mmol lϪ1 to motor giant axon close to its junction with the inner nerve ring and an extracellular electrode (R ) is shown on the inner nerve ring at the reduce the spread of epithelial depolarizations. Their efficacy L junction with the motor giant axon. Stimulating electrodes are situated as gap junction blockers (Johnston et al. 1980) was checked on the inner nerve ring further along the margin. Total length of on the exumbrellar epithelium of A. digitale, a nerve-free preparation, 0.6 cm. (B) A shock (asterisk) on the inner nerve ring Ϫ1 conducting epithelium (Mackie, 1980). At 0.5 mmol l , evokes a sequence of pulses and a slow wave (P, R and W) recorded octanol and heptanol both blocked conduction in this by the extracellular electrode (lower trace). The intracellular electrode epithelium but had no effect on the production or propagation in the motor giant axon shows a calcium spike generated by a slow of sodium spikes in the motor giant axons. postsynaptic potential coincident with the P wave (upper trace). (C) Spontaneous PÐRÐW sequence obtained after treatment with 84 mmol lϪ1 Mg2+ which has blocked the calcium spike, leaving the Results postsynaptic potential (fictive slow swimming). General characteristics of the relay system Spontaneous slow swimming is driven by pacemaker swimming’; upper trace in Fig. 2C). Thus, R and W events neurones in the inner nerve ring (Meech and Mackie, 1995). follow P impulses regardless of whether the motor giants and Extracellular recordings from the nerve ring during slow swimming muscles are actually excited and regardless of swimming reveal trains of pacemaker impulses and, as shown whether the pacemaker system fired spontaneously (Fig. 2C) in Fig. 2, similar events can be obtained by stimulating the or as a result of stimulation (Fig. 2B). There was no trace of inner nerve ring electrically. In recordings where stimulation postsynaptic events in the motor giants related to the R and W of the inner nerve ring evokes a single pacemaker impulse (P), events. We will show that the W event represents a the extracellular recording (lower trace) shows that the P depolarization of the ring giant axon and of the epithelial cells impulse is followed by a second event representing an impulse ensheathing the outer nerve ring, while the R impulse arises in in the relay system (R). This is invariably followed by a slowly a system which relays excitation from the pacemaker neurones rising and falling, negative-going after-potential, typically to the ring giant. lasting 50Ð70 ms, termed the slow wave (W). The sequence In certain stimulating positions and with rather weak shocks, PÐRÐW was seen consistently both in preparations bathed in it was sometimes possible to obtain propagated RÐW sea water (where the P impulse evoked a slow postsynaptic sequences without preceding P impulses. In such recordings, potential and calcium spike in the motor giant axons; upper the R event conducted very slowly. Although conduction trace in Fig. 2B) and in solutions containing excess Mg2+, velocity increased with repetitive stimulation, it never which suppresses the calcium spike, exposing the slow exceeded 10 cm sϪ1. In Fig. 3A, with stimulation at 0.4 Hz, the postsynaptic potential (PSP) that generated it (‘fictive slow R event evoked by the first shock conducted at 2.9 cm sϪ1 2264 G. O. MACKIE AND R. W. MEECH

A event, the R impulse followed the P event within 25Ð40 ms regardless of the length of the conducting pathway. In Fig. 3B, the conduction pathway was 11.7 mm in the upper recording and 6 mm in the lower one, but the PÐR interval was virtually the same in both (about 30 ms). Thus, the relay system was not 400 µV only triggered by the P event but was carried along by it at a 40 ms velocity considerably greater than its velocity when propagating alone. A similar case, where events generated in * R W a slowly conducting system somehow ‘hitch a ride’ on the back of a faster system, has been termed the ‘piggyback effect’ B (Mackie, 1976b). Relay (R) events can be piggybacked not only by P impulses but also by ring giant impulses (Fig. 3C), where the effect is very pronounced. In Fig. 3C, the piggybacked conduction velocity of the R event was 24 cm sϪ1 in the upper recording (P-assisted) and 41 cm sϪ1 in the lower one (ring-giant- assisted). Piggybacking by the ring giant system was not * blocked by 0.5 mmol lϪ1 octanol (Fig. 3C), 6.0 mmol lϪ1 Mn2+ P R W 200 µV or 129 mmol lϪ1 Mg2+ (not shown), suggesting that it may be mediated neither by gap junctions nor by chemical synapses. 20 ms Three-step depolarization of the ring giant axon C Following electrical stimulation of the inner nerve ring, PÐRÐW sequences in the extracellular record can be matched with depolarizations in the ring giant axon (Fig. 4A). Both P P R W and R impulses produce small postsynaptic potentials (PSPs) and, as the two normally fire in sequence, their PSPs sum to depolarise the ring giant axon by up to 16 mV. These summed PSPs together with similar depolarizations of the epithelial * R W 200 µV G cells around the outer nerve ring (see below) are responsible for the W component of the extracellularly recorded PÐRÐW 20 ms sequence. Similar two-step summing depolarizations are seen Fig. 3. Extracellular recordings made with a single recording during spontaneous fictive swimming. As shown in Fig. 4B, electrode placed on the inner nerve ring following shocks (asterisk) the first depolarization of the ring giant axon coincides with further along the margin. (A) Small shocks that fail to excite the the slow PSP recorded in the motor giant, both being due to pacemaker system evoke a propagated event consisting of a coupled pacemaker input. The second step, representing relay input, pulse (R) and slow wave (W) whose conduction velocity increases follows after an interval determined by the PÐR piggyback with repetitive stimulation (upper trace; see text). Small events relationship. following the W depolarization here and elsewhere are miniature Under our recording conditions, the two-step synaptic potentials. (B) PÐRÐW sequences evoked by shocks at depolarization of the ring giant axon due to summed different distances from the recording electrode; the PÐR latency pacemaker and relay inputs was never sufficient to bring the remains constant. (C) A shock evokes a typical PÐRÐW sequence ring giant to spike threshold. Spiking often occurred, (upper trace) while a slightly stronger shock (lower trace) fires the ring giant axon (G) as well, causing a burst of three spikes to overlap however, during PÐRÐW sequences, the spikes typically the P event. The arrival time of the RÐW sequence is determined by occurring on or near the peak of the W event. Analysis of the piggyback relationship. Octanol (0.5 mmol lϪ1) was used to reduce such cases shows that a third system, the ‘carrier’ (C) system, ring giant after-potentials that would otherwise obscure the P event. contributed to the depolarization of the ring giant axon, bringing the total change to about 20 mV (Fig. 4C,D). Impulses in the carrier system were presumably triggered by (lower trace) while the one following the fourth shock impulses in the relay system, as a carrier impulse was never conducted at 5.7 cm sϪ1 (upper trace). seen without a preceding R event. We show in the companion paper (Mackie and Meech, 1995) that the carrier system Piggyback effect normally fires in fairly close synchrony with the ring giant Although capable of propagating on its own, the relay system, producing composite signals in extracellular system has never been observed to be spontaneously active but recordings in which the two components cannot be separated. is apparently always triggered by direct stimulation of the In intracellular recordings such as Fig. 4D, however, the margin or by activity in other systems. When triggered by a P carrier postsynaptic potential could be clearly distinguished The relay system in a jellyfish 2265 as a separate entity, labelled as step 3. Production of ring in the slow tentacle system excite the longitudinal muscle of the giant spikes during slow swimming always required three tentacle, causing slow graded contractions. The postsynaptic depolarization steps, representing input from three muscle potential is seen as a large upward after-potential presynaptic systems, the pacemaker, relay and carrier following the smaller presynaptic nervous event. What sets off systems, firing in that order. Carrier pulses were never seen the slow tentacle pulse is not definitely known, but triggering without a preceding R pulse, nor R pulses without a P event, probably involves the relay system, as P impulses alone failed suggesting that the PÐRÐC cascade represents triggering of to excite a slow tentacle pulse and triggering often occurred in the relay system by the pacemaker and of the carrier system the absence of C events. In traces where the ring giant axon fired by the relay. following a three-step depolarization (as described above), the Although, as noted above, ring giant spikes generated in this ring giant spike was followed by an impulse in the tentacle giant way usually occurred on or near the peak of the W event, they axon, which caused a twitch contraction of the tentacle. This is sometimes occurred earlier or later. The time relationships shown in Fig. 5C, where the tentacle giant spike was presumably depend on (a) the PÐR triggering interval, (b) the superimposed on the slow tentacle pulse and its after-potential. RÐC triggering interval and (c) the distance from the recording Thus, activity in the pacemaker neurones can be relayed by the electrode at which spike threshold was first reached. We are R system to the tentacles during slow swimming, causing both dealing with a linear preparation representing the cut and graded and twitch-type contractions. straightened margin of a circular animal. Several systems run along the strip in parallel and interact at numerous points along A their length. A ring giant spike, once initiated, can presumably 2 travel at its own high conduction velocity and, by the time it reaches the recording electrode, can catch up with or even 1 overtake events in the presynaptic systems that triggered it. Again, with impulses generated by a stimulating electrode near 10 mV the recording site and propagating both towards and away from it, the ring giant may fail to reach spike threshold except at a R 100 µV point remote from the recording electrode, allowing impulses * P 20 ms in the presynaptic systems to arrive at the recording electrode well in advance of the ring giant spike, despite the more rapid B conduction velocity of the latter. 20 mV Triggering of tentacular contractions 2 PÐRÐW sequences in the marginal nerves were typically 40 ms 5mV followed by electrical events in the system previously referred to as the ‘small tentacle pulse system’ (Bickell-Page and Mackie, 1 1991), and here termed the ‘slow tentacle conduction system’ or ‘slow tentacle system’. A typical example of a pulse in this C 20 mV system is shown in Fig. 5B labelled TS (upper trace). Impulses

200 µV Fig. 4. (A) A shock (asterisk) on the inner nerve ring evokes a typical 40 ms 3 PÐRÐW sequence recorded extracellularly (lower trace). A 2 simultaneous intracellular recording from the ring giant axon (upper 1 trace) shows depolarizations due to pacemaker (1) and relay (2) system inputs. (B) Two superimposed traces showing simultaneous intracellular recordings from a motor giant axon (upper trace) and an adjacent ring giant axon (lower trace) during a PÐRÐW sequence (not P R shown, but as in A). The slow postsynaptic potential (PSP) in the D motor giant corresponds in time to depolarization 1 in the ring giant, both being due to pacemaker input. (C) Two successive sweeps superimposed showing a three-step depolarization of the ring giant axon (upper trace) representing successive input from the pacemaker, relay and carrier systems and culminating in a ring giant spike (3) on the second sweep. The extracellular correlate of the carrier system is lost within the complex deﬂection representing the ring giant spike and its after-potential (lower trace). (D) Intracellular recordings from 20 mV the ring giant axon at the peak of three-step depolarizations. Supra- and subthreshold sweeps are superimposed, showing the production 10 ms of a ring giant spike by a carrier system postsynaptic potential (3). 2 3 2266 G. O. MACKIE AND R. W. MEECH

Outer nerve ring Fig. 5. (A) Drawing to show a preparation pinned with its exumbrellar side up. Most of the tentacles have been truncated, leaving one (left-hand side) intact. An RU extracellular recording electrode (RU) is shown attached to it. A second extracellular electrode (RL) is sited on the Stimulus outer nerve ring close to the tentacle. Stimuli are RL delivered on the outer nerve ring further along the margin. (B) Extracellular recordings from a tentacle (upper trace) following shocks (asterisk) on the margin evoking BC PÐRÐW sequences, recorded from the outer nerve ring adjacent to the tentacle (lower trace). A slow tentacle pulse (TS, with its large upward after-potential) was triggered by the R event. (C) As for B except that a ring * TS * giant spike (G) occurred following the R event and 200 µV resulted in a tentacle giant impulse in the tentacle, seen superimposed on the slow tentacle pulse and its upward 40 ms after-potential. P R W G

Epithelial depolarizations nervous sub-systems. They were frequently seen to be blocked Attempts to penetrate the ring giant often gave an initial for several seconds by spikes in the ring giant axon (Fig. 6B) resting potential of Ϫ72 to Ϫ75 mV. In such cases, the and they appeared to be induced by both spontaneous (Fig. 6C) electrode tip appeared to be located in the epithelium covering and shock-induced PÐRÐW sequences (Fig. 6D). In several the ring giant rather than in the ring giant itself, and this was recordings, shocks delivered more than 500 ms after a verified by ionophoresis of carboxyfluorescein. The dye spread spontaneous MSP evoked a PÐRÐW sequence that consistently diffusely through a patch of epithelium around the injection generated an MSP on the descending slope of the W point. During these intracellular recordings, the epithelium was component. observed neither to spike nor to conduct action potentials Intracellular recordings from epithelial cells show that the although it showed a great deal of spontaneous background latter depolarize during each RÐW event. In fact, as noted activity, presumably reflecting activity in the neurones it earlier, the W event is in part the extracellular correlate of this ensheaths. The spontaneous potentials showed no clear effector depolarization. In Fig. 7A, an RÐW sequence piggybacked by correlates and were strictly local, although two electrodes a shock-evoked ring giant event corresponds in time to a 17 mV placed close together could pick up the same events (Fig. 6A). long-lasting depolarization of the epithelium. The extracellular Though not truly ‘miniature’ events (because they reached record captured only the early part of this depolarization owing 5 mV in intracellular and 150 ␮V in extracellular recordings), to capacitative coupling in the amplifiers. The epithelial cell we tentatively regard them as the equivalent of miniature end- was also depolarized following a simple ring giant spike plate potentials and suggest that they represent spontaneous (Fig. 7B). In Fig. 7C, stimulation evoked a PÐRÐW sequence transmitter release by the ensheathed neurones. Similar events where the W event again corresponded to a 17 mV recorded in the subumbrellar myoepithelium were termed depolarization in the epithelium. In the same preparation ‘miniature synaptic potentials’ by Kerfoot et al. (1985), a term (Fig. 7D), a PÐRÐWÐring giant sequence also occurred, and a we adopt here. Miniature synaptic potentials (MSPs) usually PSP due to the ring giant event is seen mounted on the top of occurred in fairly regular patterns, although phases of high the epithelial depolarization. activity (approximately 3 Hz) often alternated with phases of low activity (approximately 1 Hz). Similar patterns of MSPs were also seen in the pad cell, a specialized epithelial cell Discussion associated with hair cell clusters (Arkett et al. 1988; Mackie, The findings reported here show that the swimming 1989b). Like other epithelial cells showing MSPs, the pad cell pacemaker neurones, whose most obvious function is to fire contributes to the ensheathment of the outer nerve ring. the motor giant axons in the slow swimming response, also Patterns of MSPs can be affected by events in at least two produce PSPs in the ring giant axon and trigger events in The relay system in a jellyfish 2267

A A

* 5s

G R W

B B

G 5s 20 mV G C 200 µV 40 ms

C 1s

400 µV *RWM * RW M 40 ms D 10 mV Fig. 6. Miniature synaptic potentials (MSPs). (A) Simultaneous intracellular (upper trace) and nearby extracellular (lower trace) 100 ms recordings of spontaneous MSP patterns showing alternating high- 200 µV and low-activity phases. (B) In an extracellular recording, a regular pattern of MSPs is blocked for several seconds following a burst of spikes in the ring giant axon (G). (C) In an extracellular recording from the epithelium, spontaneous PÐRÐW sequences (dots) occurring during rhythmic MSP activity induce precocious MSPs, briefly altering the rhythm. (D) Induction of MSPs (M) following shock- induced RÐW sequences (shock artefact marked with an asterisk); two G M similar extracellular recordings on an expanded time scale (see also * Fig. 3A). Fig. 7. Intracellular recordings from the epithelium overlying the outer nerve ring (upper traces) during nervous activity recorded another system of interneurones, the relay system. The relay extracellularly in the same vicinity (lower traces). (A) A shock- system similarly produces PSPs in the ring giant and triggers evoked ring giant spike (G) and piggybacked RÐW sequence are both events in a third system, the carrier system, which in turn accompanied by epithelial depolarizations. In B, the ring-giant- generates PSPs in the ring giant. The result is a sequential, evoked depolarization is seen alone. In C, a PÐRÐW sequence (the P three-step depolarization of the ring giant that may bring following very closely upon the shock artefact, marked with an asterisk) gives rise to an epithelial depolarization corresponding to the it to spike threshold. These interactions are shown W event. In D, epithelial depolarizations are associated with both W diagrammatically in Fig. 8. and ring giant events (G) in a PÐRÐWÐring giant sequence. In C and Whenever the ring giant fires, the tentacles contract together D, some miniature synaptic potentials events (M) are seen on both in a twitch response. Thus, twitch contractions of the tentacles channels, but the two electrodes were not close enough for both frequently accompany slow swimming even though we think electrodes to pick up the same events all the time. of such contractions as ‘belonging’ to the escape response. By virtue of the relay and carrier systems, the animal has bridged the gap between its escape and non-escape circuits and can depolarization cascade described here may be necessary to utilize an escape component in its normal, slow swimming bring it to spike threshold, given its presumed high membrane behaviour. capacitance. The pacemakerÐrelayÐcarrier triggering The ring giant is an enormous axon, and the three-step sequence can therefore be seen as a way of amplifying the 2268 G. O. MACKIE AND R. W. MEECH initial pacemaker impulse to the level needed to fire the ring swimming bout. The firing frequency of the tentacle system giant. triggered through the relay pathway cannot exceed the In Fig. 8 we show the ring giant and carrier system as lying frequency of slow swimming, which is only about 0.5 Hz and, immediately adjacent to one another, connected by given the incremental, graded nature of the tonic response, bidirectional excitatory links and with the various inputs and such a low frequency would probably be insufficient to outputs entering at or exiting from the interface between the contract the tentacles early enough and strongly enough to two. The reason for this ambivalence is that the two systems prevent excessive hydrodynamic drag or the danger of usually fire in synchrony and it is hard to know which system entanglement. We suggest, therefore, that the twitch is primarily involved in any given interaction. This question is contractions serve to bring about a rapid, high degree of further addressed by Mackie and Meech (1995). contraction which can then be maintained tonically by the Twitch contractions of the tentacles, mediated by the graded input. Here, however, physiological speculation has tentacle giant axon, are not the only way in which the tentacles gone some way ahead of behavioural observations and there is can contract. The animal has a mechanism for graded, tonic a need to go back to the animal in nature and observe its contractions that is also activated during slow swimming. At behaviour more closely. each slow swimming event, the relay system directly excites If, following a pacemaker impulse, the ring giant axon is the slow tentacle system (Fig. 8). The tentacle longitudinal depolarized in a three-step cascade and reaches spike muscles contract incrementally with each slow tentacle pulse. threshold, the resulting action potential should not only cause It might seem that twitch contractions of the tentacles are the tentacles to twitch but should also fire the motor giant redundant, given the existence of tonic contractions, but it axons, causing sodium spikes in them and leading to escape seems likely that the twitch contractions are needed to bring swimming. This does not happen, however, and it appears that about the requisite degree of shortening at the start of a slow the ring giant axon to motor giant axon excitation pathway is delicately adjusted so that the occurrence of a slow swimming event, whether real or fictive, somehow blocks transmission Motor giant axon from the ring giant to the motor giant or prevents the generation of a sodium spike. How it does so is unknown at present. Motor giant rootlet We know of no other case where a giant axon mediating escape behaviour can be ‘borrowed’ to augment a non-escape Pacemaker system activity. The tail-flip response of crayfish can be mediated by giant axons or by non-giants. There are separate pathways to Relay system the muscles in the two cases, and one pathway (the fast flexor Carrier system motor neurones) is used in both responses. In addition, certain non-giant interneurones that innervate the fast flexor motor Ring giant axon neurones in non-giant tail-flipping are ‘commandeered’ during giant-mediated flips (Krasne and Wine, 1984). These examples of shared circuitry recall the situation in A. digitale in so far as Tentacle giant axon the motor giant axons are concerned, for the latter provide the final common pathway for both escape and non-escape swimming, but the crayfish offers no counterpart to the commandeering by A. digitale of its ring giant axons during non-escape locomotion.

The piggyback effect Our findings show that the relay system is capable of Tentacle slow system conducting slowly on its own but that, when impulses in it are preceded by impulses in the pacemaker system or the ring giant axon, it shows a much higher conduction velocity. The mechanism for this ‘piggyback’ effect (shown as crooked Excitatory Piggyback synaptic effect arrows in Fig. 8) is unclear, but the drug effects reported here input suggest that piggybacking may not be due to conventional junctional transmission but may involve extracellular spread of Bidirectional action currents between faster-conducting and slower- input conducting elements. Such ‘field effects’ are, as noted by Fig. 8. Diagrammatic ‘side-on’ view of the central nervous system Bullock (1984), hard to prove or disprove, and a lengthy summarizing the interactions discussed in the text. There are three discussion of the problem would be inappropriate at this kinds of synaptic connection. Some have an excitatory synaptic input, juncture, but it should be noted that cnidarian neurones are not some are bidirectional and some show the piggyback effect. individually ensheathed (Horridge et al. 1962). As shown for The relay system in a jellyfish 2269

Polyorchis (Spencer and Arkett, 1984), hydrozoan conduction Meech, 1995) are not picked up in the epithelium. Processes systems consist of clusters of neurones functioning more or of the epithelial cells ensheath bundles of neurones and could less in synchrony and lying close to, and probably function actively in distributing electrical field effects or intermingling with, other such clusters. In A. digitale, epithelial passively as insulating barriers. Thus, they could play some processes do appear to subdivide the nerve rings into separate part in mediating piggyback interactions between neuronal territories to some extent, but the evidence for this is tentative, subsets. and it would not be surprising if cross-talk between different systems sometimes occurred. Indeed, Spencer (1981) has Comparisons with other medusae suggested that inhibition of the swimming motor neurones in We noted earlier that the relay system has no obvious Polyorchis penicillatus may be due to field effects from counterpart in other medusae. Coordination of the tentacles surrounding epithelial cells. Whatever the mechanism, in other medusae is carried out by a system originally termed piggybacking of relay system impulses by the pacemaker the marginal pulse system (Passano, 1965) and represented in system is clearly of functional importance because it ensures Polyorchis penicillatus by the ‘B’ system (Spencer, 1978; that relay impulses follow pacemaker impulses within the Spencer and Arkett, 1984; Arkett and Spencer, 1986). This fairly short time interval necessary for summing of their system runs around the margin in the outer nerve ring and postsynaptic potentials in the ring giant during PÐRÐC down each tentacle. Its rate of firing determines the degree of sequences. Piggybacking of relay impulses by ring giant spikes tentacular contraction. It is excited when the tentacles are is not so clearly advantageous. Its effect would be to prolong stimulated, but also fires spontaneously, the pattern of firing the state of tentacle contraction slightly. being the same all round the margin. Bursts of B pulses are When no other systems are excited and the relay system is seen prior to swimming, so that by the time swimming starts, stimulated with a series of shocks, its conduction velocity the tentacles are already contracted. The B system receives increases following the first shock. A similar phenomenon, input from the ocelli in the form of excitatory PSPs following termed ‘action potential facilitation’, has been described in shadowing of the ocelli and responds with bursts of spikes. oligochaetes (Bullock, 1951; Drewes et al. 1978), in which a These in turn produce excitatory PSPs in the swimming second impulse following the first within about 6 ms shows a motor neurones (equivalent to the pacemaker system of velocity increase of up to 20 %. In A. digitale, the velocity A. digitale) and in the epithelium covering the outer nerve increase could, like the piggyback effect, be a phenomenon in ring. the field effect category if stimulating or action currents are The relay system of A. digitale, like the B system, runs stored capacitatively in nearby excitable tissues long enough around the margin, but it does not run down the tentacles. to affect the excitability level of the relay system at its next Tentacle posture is controlled by another system, the slow impulse. tentacle system, which is fired by the relay system. There is no evidence for transmission in the reverse direction. Tentacle Interactions between nerves and epithelial cells stimulation produces bursts of tentacle pulses that remain The origin and functional significance of the events recorded localized within that tentacle and do not enter the relay system from the epithelial cells ensheathing the neurones of the outer or spread around the margin to other tentacles. Unlike the B nerve ring are largely unknown at present, but some system, the relay system is not spontaneously active and it does understanding of their basis is essential for the interpretation not fire prior to swimming. On the contrary, impulses in it are of the somewhat complicated extracellular recordings from the triggered by the same events that produce swimming. It shows margin. If the events interpreted as miniature synaptic no response to shadowing or illumination and although R potentials (MSPs) do indeed represent spontaneous transmitter impulses evoke epithelial depolarizations (W events) there is release, neuro-epithelial synapses must occur, but as yet we no effect on the pacemaker system. If A. digitale has a have no ultrastructural evidence of such junctions in the counterpart to the B system, it could only be the slow tentacle immediate vicinity of the outer nerve ring. The blocking of system, but this would require that the system has become MSP patterns by ring giant spikes (Fig. 6B) suggests that restricted to each tentacle rather than interconnecting all the carrier neurones may be the source of MSPs, as they fire when tentacles. the ring giant fires, which would deplete their vesicular It would therefore appear that the relay system has evolved content. Gap junctions occur between the ring giant and in conjunction with escape circuitry in rhopalonematid surrounding epithelial cells (Mackie, 1989b) and might medusae as a mechanism for bringing about concerted mediate the depolarization of the epithelium by ring giant tentacle contractions during slow swimming. If this function spikes. Their presence might also help to account for the was originally performed by the B system, the latter has either observation that the ring giant and the enveloping epithelial been lost or has become restricted to each tentacle forming cells depolarize in synchrony following R impulses, producing the tentacle system. The relay system coordinates the the W potential. If the two systems are coupled, however, the tentacles in two ways: (a) by directly exciting the tentacle coupling must be fairly weak because MSPs are recorded only system, producing tonic contractions, and (b) by helping to in the epithelium, and the background oscillations and synaptic depolarize the ring giant axon to spike threshold, the resulting activity characteristic of the ring giant axon (Mackie and ring giant spikes causing excitation of the tentacle giant axons 2270 G. O. MACKIE AND R. W. MEECH and twitch contractions. Contraction of the tentacles during MACKIE, G. O. (1975). Neurobiology of Stomotoca. II. Pacemakers swimming is clearly advantageous in terms of drag reduction and conduction pathways. J. Neurobiol. 6, 357Ð378. and because it prevents tangling. It is doubtful if the MACKIE, G. O. (1976a). Propagated spikes and secretion in a necessary degree of contraction could be achieved by coelenterate glandular epithelium. J. gen. Physiol. 68, 313Ð325. activating the slow tentacle system alone as the relay input MACKIE, G. O. (1976b). The control of fast and slow muscle frequency cannot exceed the rather low value set by the slow contractions in the siphonophore stem. In Coelenterate Ecology and Behavior (ed. G. O. Mackie), pp. 647Ð659. New York: Plenum. swimming pacemaker system. The animal has therefore made MACKIE, G. O. (1980). Slow swimming and cyclical ‘fishing’ behavior use of part of its escape circuitry (the ring giant and tentacle in Aglantha digitale (Hydromedusae: Trachylina). Can. J. Fish. giants) to enhance tentacle contractions during its normal aquat. Sci. 37, 1550Ð1556. slow swimming behaviour. MACKIE, G. O. (1989a). Louis Agassiz and the discovery of the coelenterate nervous system. Hist. Phil. Life Sci. 11, 71Ð81. We thank Professor A. O. D. Willows, Director of the Friday MACKIE, G. O. (1989b). Evolution of cnidarian giant axons. In Harbor Laboratories, for providing space and superb facilities, Evolution of the First Nervous Systems (ed. P. A. V. Anderson), pp. Linda Bédard for the drawings used in Fig. 1 and Matthew 395Ð407. New York: Plenum. Meech for Fig. 8. G.O.M. acknowledges the support of the MACKIE, G. O. AND MEECH, R. W. (1985). Separate sodium and Natural Sciences and Engineering Research Council of Canada calcium spikes in the same axon. Nature 313, 791Ð793. MACKIE, G. O. AND MEECH, R. W. (1995). Central circuitry in the for operating and equipment funding. R.W.M. thanks the jellyfish Aglantha digitale. II. The ring giant and carrier systems. Wellcome Trust and the Invertebrate Neuroscience Initiative J. exp. Biol. 198, 2271Ð2278. of the Science and Engineering Research Council of the UK MACKIE, G. O., NIELSEN, C. AND SINGLA, C. L. (1989). The tentacle for travel funds. cilia of Aglantha digitale (Hydrozoa: Trachylina) and their control. Acta zool. 70, 133Ð141. MACKIE, G. O. AND SINGLA, C. L. (1975). Neurobiology of Stomotoca. References I. Action systems. J. Neurobiol. 6, 339Ð356. ARKETT, S. A., MACKIE, G. O. AND MEECH, R. W. (1988). Hair cell MEECH, R. W. AND MACKIE, G. O. (1993a). Ionic currents in giant mechanoreception in the jellyfish Aglantha digitale. J. exp. Biol. motor axons of the jellyfish, Aglantha digitale. J. Neurophysiol. 69, 135, 329Ð342. 884Ð893. ARKETT, S. A. AND SPENCER, A. N. (1986). Neuronal mechanisms of MEECH, R. W. AND MACKIE, G. O. (1993b). Potassium channel family a hydromedusan shadow reflex. J. comp. Physiol. A 159, 201Ð213. in giant motor axons of Aglantha digitale. J. Neurophysiol. 69, BICKELL-PAGE, L. R. AND MACKIE, G. O. (1991). Tentacle autotomy 894Ð901. in the hydromedusa Aglantha digitale (Cnidaria): an ultrastructural MEECH, R. W. AND MACKIE, G. O. (1995). Synaptic potentials and and neurophysiological analysis. Phil. Trans. R. Soc. Lond. B 331, threshold currents underlying spike production in motor giant axons 155Ð170. of Aglantha digitale. J. Neurophysiol. (in press) BULLOCK, T. H. (1951). Facilitation of conduction rate in single nerve MILLS, C. E., MACKIE, G. O. AND SINGLA, C. L. (1985). Giant nerve fibres. J. Physiol., Lond. 114, 89Ð97. axons and escape swimming in Amphogona apicata with notes on BULLOCK, T. H. (1984). Comparative neuroethology of startle, rapid other hydromedusae. Can. J. Zool. 63, 2221Ð2224. escape and giant fiber-mediated responses. In Neural Mechanisms PASSANO, L. M. (1965). Pacemakers and activity patterns in medusae: of Startle Behavior (ed. R. C. Eaton), pp. 1Ð13. New York: Plenum Homage to Romanes. Am. Zool. 5, 465Ð481. Press. ROBERTS, A. AND MACKIE, G. O. (1980). The giant axon escape system DREWES, C. D., LANDA, K. B. AND MCFALL, J. L. (1978). Giant nerve of a hydrozoan medusa, Aglantha digitale. J. exp. Biol. 84, activity in intact, freely moving earthworms. J. exp. Biol. 72, 303Ð318. 217Ð227. ROMANES, G. J. (1876). Preliminary observations on the locomotor HORRIDGE, G. A. AND MACKAY, B. (1962). Naked axons and system of medusae. Phil. Trans. R. Soc. Lond. B 166, 269Ð313. symmetrical synapses in coelenterates. Q. Jl microsc. Sci. 103, SPENCER, A. N. (1978). Neurobiology of Polyorchis. I. Function of 531Ð541. effector systems. J. Neurobiol. 9, 143Ð157. JOHNSTON, M. F., SIMON, S. A. AND RAMÓN, F. (1980). Interaction of SPENCER, A. N. (1979). Neurobiology of Polyorchis. II. Structure of anaesthetics with electrical synapses. Nature 286, 498Ð500. effector systems. J. Neurobiol. 10, 95Ð117. KERFOOT, P. A. H., MACKIE, G. O., MEECH, R. W., ROBERTS, A. AND SPENCER, A. N. (1981). The parameters and properties of a group of SINGLA, C. L. (1985). Neuromuscular transmission in the jellyfish electrically coupled neurones in the central nervous system of a Aglantha digitale. J. exp. Biol. 116, 1Ð25. hydrozoan jellyfish. J. exp. Biol. 93, 33Ð50. KRASNE, F. B. AND WINE, J. W. (1984). The production of crayfish SPENCER, A. N. AND ARKETT, S. A. (1984). Radial symmetry and the tailflip escape responses. In Neural Mechanisms of Startle Behavior organization of central neurones in a hydrozoan jellyfish. J. exp. (ed. R. C. Eaton), pp. 179Ð211. New York: Plenum Press. Biol. 110, 69Ð90.